Start a new topic

Automatic insertion of exact matches that only differ in a trailing space

Would it be possible to enhance Translate > Insert all EMs in such a way that EMs that only miss a trailing space, will be inserted too?


Same for CMs.

There is much more room:


There is the following number of blanks here:

There is the following number of blanks here_:


The possibility of ignoring blanks before sentence marks (some French love them, others not) would be nice.


Maybe this programming could even become obsolete by introducing a new segment state (pretranslated) and sorting the matches by match value (mantra anybody?).


1 person likes this

An alternative (temporarily?) approach would be to:

  • Clone your memory.
  • Remove all trailing spaces (source, target) in the cloned memory.
  • Merge this cloned memory with the originating memory.
Only useful for really big projects?

> Only useful for really big projects.


Only useful for really big projects, indeed, as it is quite common in French texts. Same for 


« This article describes the common bla bla bla.

«This article describes the common bla bla bla.


No EM, no CM inserted.

> Automatic insertion of exact matches that only differ in a trailing space


CafeTran can insert such imperfect exact matches automatically as you enter a given segment. As you need to correct anyway (e.g erasing the trailing space), there is not much difference whether the match is inserted beforehand or "as you go". The second method is even safer because you might have missed those slight errors via the automatic insertion in one batch.

> as you enter a given segment


Depending on TM size, this can make a considerable difference. Even when using preliminary matching (seriiously with medium TMs > 100 MB), and with much RAM. 


Just as a note: For time management and workflow there is a difference if you need to watch CT working some seconds until the results comes out and go on or if you let the machine work, use the time in other productive manner and review the results afterwards. Come on, this is what computers are for. 


As said several times before, there are some really common scenarios where CT fails. One of them is this here, with big TMs and very similar segments (only differentiated by a space).

With Total Recall and/or Preliminary matching, you can work with huge TMs just fine. I disagree with you on this, and that's perfectly fine.

Login to post a comment